Evaluation of driver's eco-driving skills based on fuzzy logic model: A realistic example of vehicle operation in real-world conditions

Abstract

It is well known that fuzzy logic is a processing tool in circumstances lacking of clear linguistic information, as well as making conclusions based on imprecise assertions and rough data. Eco-driving rules that the drivers should comply with are not always made of concrete values (exact acceleration / deceleration rates, torque or headway / distance kept from the vehicle ahead, etc.), but often linguistically expressed and subjective (e.g. soft acceleration, mid-range engine speed, soft deceleration, suffi cient distance, etc.). Therefore, the authors recognized fuzzy logic potentials as an effi cient tool to overcome all mentioned barriers and thus to increase vehicle energy efi ciency and reduce emissions of harmful gases which are main goals of eco-driving. The primary objective of this paper is to raise the awareness on the potentials and effi ciency of fuzzy logic systems' use in eco-driving as a tool for achieving more ecologically & economically sustainable road transport. The rules that drivers should follow in order to achieve and maintain eco-driving goals, as well as the parameters to be monitored to evaluatedriver's behaviour i.e. the compliance with eco-driving rules are presented in the paper. The authors propose a driver rating system based on the fuzzy logic model constructed within MatLab. Within the proposed model the input parameters are actual acceleration/deceleration rates, engine speed and accelerator pedal pressure(APP) and while model output are driver ratings (scores ranging from 0 to 10 points) after completed a driving cycle. A real-world example based on data collected via vehicle OBDII connector by a TEXA logging device in realistic vehicle operation conditions. The consequent actual results of drivers' behaviour rating tool based on the proposed model are presented in the paper.

Assessment of the driver's driving style using supervised machine learning

“…Proces mašinskog učenja sastoji se od faza: priprema podataka, treniranje modela, validacija modela, testiranje modela i primena modela, tj. predikcija ciljne varijable [6]. Predikcija ocene vozačevog stila vožnje u svakoj sekundi vožnje realizovana je kroz sve gore navedene faze procesa mašinskog učenja, posebno za svakog od šest različitih vozača, na tri izabrane rute, pri čemu je svaki vozač na svakoj ruti upravljao istim vozilom tokom pet dana u nedelji.…”

“…Stil vožnje vozača je jedan od faktora koji ima značajan uticaj na potrošnju goriva i na energetsku efikasnost vozila [6]. Ponašanje vozača je veoma složen Adresa autora: Slađana Janković, Univerzitet u Beogradu, Saobraćajni fakultet, Beograd, Vojvode Stepe 305 e-mail: s.jankovic@sf.bg.ac.rs Rad primljen: 07.02.2022.…”

“…Autori u radu [6] razvili su model fazi logike (engl. Fuzzy Logic) za dobijanje ocene vozačevog stila vožnje kod putničkih automobila, pri čemu su utvrdili da što je ocena vozača bolja to je specifična potrošnja goriva (u l/100 km) njihovih vozila manja, i obrnuto.…”

Evaluation of the influence of terrain and traffic road conditions on the driver’s driving performances by applying machine learning

“…In order to achieve the longterm effects of "eco-driving" training courses, too, there is a need to supervise the driver's driving style through the constant supervision of the values of the vehicle parameters that influence fuel consumption [14]. In the paper [15], the authors scored the driver's driving style in passenger cars by using the following vehicle parameters: the engine speed, the vehicle acceleration/deceleration and the accelerator pedal position. They determined that the higher the score for the driver (a passive driving style), the smaller specific fuel consumption (in l/100 km), and vice versa.…”